Agent-based modeling of land-use changes and vulnerability assessment in a coupled socio-ecological system in the coastal zone of Sri Lanka

In recent decades, the different impacts of global change have led to an increased exposure and vulnerability of coupled socio-ecological systems (SES) to various disturbances and stressors. Human-induced land-use and land-cover changes (LUCC) are an important factor contributing to the vulnerability of such systems. It is therefore reasonable to analyse vulnerability and LUCC in a combined effort. In this study, projections for future LUCC around an estuary in southwestern Sri Lanka were developed. In a subsequent step, the vulnerability of the coupled coastal system to natural hazards was analyzed based on the effects of the tsunami in December 2004. Changes in land use and land cover are the most important impacts of human activities on the earth system, with mostly negative consequences for both ecosystems and social systems. Awareness has grown about the complexity of the driving forces of LUCC, and various modeling approaches have been developed for generating scenarios on different spatial scales. Agent-based models (ABMs) with their focus on the simulation of human decision making and the interdependencies between humans and their environment are seen as a promising and flexible approach. They are best capable of capturing the complexity of coupled SES and of providing a natural description of the systems under consideration. This study employed the Land-Use Dynamics Simulator (LUDAS), an ABM consisting of four modules, which was developed in 2005 for an upland watershed in central Vietnam. In addition to the household and the landscape modules, the policy component captures land-use related policies and other external parameters that impact decision making with respect to land use. The decision-making sub-module as the core of the model simulates the decision-making processes of the household agents. First, LUDAS was modified according to different biophysical and socio-economic circumstances in the coastal study region. Current land use was analyzed with high-resolution satellite images (Ikonos), while socio-economic information was based on extensive household surveys and field visits to agricultural plots, complemented by expert interviews on the local, regional, and national level. Principal Component Analysis and Cluster Analysis were used to classify the sampled population into five different livelihood groups. In a next step, binary logistic regression was employed to evaluate decision making with regard to land-use choices of the households within the different livelihood groups. Access of households to agricultural extension services and agricultural subsidies were identified as the most important external parameters assumed to have an impact on land use and economic patterns in the study region. Notwithstanding the impact of the policy factors, all 20-year simulations revealed slight decreases in the share of income from farming activities, while the overall income of the households of all groups remains more or less constant. Furthermore, landholdings of all households decrease in size, which is mainly due to decreases in homestead area, while agricultural land uses remain on the same level or show slight increases. Access to subsidies has a stronger effect on land use and livelihoods than access to extension services. It increases the share of income from farming activities, and modifies the development of the mixed cinnamon and paddy area. Changes in access to extension services only result in minor LUCC with large uncertainties for all scenarios. Comprehensive vulnerability approaches attempt to capture the exposure, sensitivity, and resilience of coupled SES. This study employed a guiding multi-level vulnerability framework with a specific focus on the linkages and feedbacks between social and environmental components of the system and their implications for the vulnerability of the SES. Data for the vulnerability assessment were collected through household surveys, expert interviews, analysis of secondary material, statistical analysis, and field and vegetation surveys. The assessment revealed the high vulnerability of households depending on fishery and labor. The vulnerability of the fishery households was due to a higher exposure and to failures in individual and institutional coping after the tsunami. The vegetation survey with the subsequent statistical analysis identified the varying impact of coastal vegetation on the energy of the tsunami waves.